During embryogenesis, EndoB, the mouse form of human keratin 18 (K18), is expressed in a complex spatial and temporal pattern in various embryonic epithelia. We have compared the expression of transgenic human K18 to the endogenous mouse homolog and to the coexpressed, complementary keratin 8 homolog, EndoA, during postimplantation mouse embryogenesis and fetal development in order to determine the developmental expression pattern of the human gene in a mouse environment. The tissue distribution of K18 protein was identical to that of endogenous EndoB in both 7.5- and 13.5-day-old embryos, except for certain heart, eye, and extraembryonic mesodermal tissues in which K18 was not detected. These results indicate that the 10-kb K18 gene specifies appropriate developmental expression in the mouse and support previously reported differences in K18 expression in human and mouse fetal heart. We have also compared the expression patterns of K18 to a series of constructions that utilize the Escherichia coli gene for beta-galactosidase (lacZ) as a reporter gene. Some of these constructions were regulated correctly in embryos during development of the germ layers. However, none was expressed consistently in extraembryonic or in adult tissues. Analysis with methylation-sensitive restriction enzymes revealed that hypermethylation of the CpG-rich prokaryotic reporter gene was not the cause of its silence in adult transgenic liver. However, the repressed state of K18-LacZ transgenes in adult liver was correlated with a different chromatin state that lacked diagnostic DNase hypersensitive sites found in K18 transgenic liver. Expression of the lacZ reporter gene did not accurately reflect the developmental pattern of K18 even in constructions that used all available K18 sequences. We conclude that in these contexts, the lacZ gene was not a developmentally neutral reporter gene.